The mechanism of membrane insertion of proteins has been extensively studied in both pro- and eukaryotic cells. At the MOM and ER membranes the emphasis has been on the roles of the GIP and SecYEG/Sec61 PCCs (Schatz and Dobberstein, 1996). The mechanisms of membrane targeting and insertion have been mostly studied using polypeptides that are dependent on these protein complexes for proper biogenesis.
These protein precursors contain signal peptides of presequences that direct them to the target membrane. Here, membrane targeting and membrane insertion of C-tail anchored proteins that do not have the canonical protein targeting signals, like the signal peptides or presequences, were studied using cytochrome b(5) and N-Bak as models.
7.1 Direct membrane insertion of C-tail anchored proteins at the endoplasmic reticulum membrane
The determinants for accurate membrane targeting of cytochrome b(5) have been previously characterized. The ER and mitochondrial isoforms have been shown to be specif ically targeted in a post-translational manner (Borgese et al., 2001;
D’Arrigo et al., 1993). Targeting of both proteins was determined to be dependent on the length of the TMD. MOM targeting required the presence of positive charges flanking the TMD (De Silvestris et al., 1995; Kuroda et al., 1998; Pedrazzini et al., 1996). The mechanism of membrane insertion of b(5), however, was unknown and therefore the subject of this body of work.
The results presented here indicate that b(5) does not require the Sec61 PCC to insert into the ER membrane.
In experiments in S. cerevisiae, in vivo,
the loss of functional Sec61 PCC, or the Sec62/63 targeting protein complex, did not affect the post-translational membrane insertion of b(5) (I). The same conclusion was reached with in vitro experiments, where depletion of the Sec61 complex from proteoliposomes, reconstituted from microsomal protein extracts, did not inhibit membrane insertion (II). The only protein activity that was identified to influence membrane insertion was the requirement for cytosolic chaperones (I). These proteins are proposed to maintain the cytosolic b(5) polypeptide in a membrane insertion competent state, probably by preventing inappropriate binding of the hydrophobic TMD to other proteins.
These cytosolic chaperones are therefore not required for the membrane insertion event per se. Membrane insertion was apparently independent of ATP (I). The only determinant we found to be important for membrane insertion of b(5) was a low cholesterol content in the bilayer. In both microsomes and liposomes, cholesterol concentrations that mimicked the Golgi and plasma membrane cholesterol content inhibited the accommodation the TMD of b(5). Cholesterol poor membranes however, allowed membrane insertion (II).
I will hereby refer to this PCC-independent membrane insertion mechanism as “direct membrane insertion”.
The cholesterol dependent inhibition of insertion provides a model for how membrane targeting of b(5) is achieved in vivo. Although the ER membrane is the major site of cholesterol synthesis, ER membranes only contain 0.5 % of the total cellular cholesterol, most of it residing at the plasma membrane (Lange et al., 1999; Reinhart et al., 1987). In general, the lipid composition of membranes of the
Discussion and conclusions
organelles along the secretory pathway, from the ER to the plasma membrane, changes in a steady gradient (Table 3).
Saturated species of phospholipids and cholesterol are enriched along this pathway (Schneiter et al., 1999). Thus the ER membrane, low in cholesterol and saturated lipids, is more fl uid when compared to the Golgi and plasma membranes (Le Gall et al., 2004). TA proteins, such as b(5), targeted to the ER membrane could thus only be able to directly insert into this membrane due to its characteristic fl uidity.
Thus, the post-ER membranes have to be reached subsequent to targeting to the ER membrane, via vesicular transport along the secretory pathway.
Membrane insertion of TA proteins into the ER membrane is not the only protein transport mechanism “regulated”
by proper cholesterol content. Cholesterol is essential for the the maintenance of ER-to-Golgi transport of secretory proteins (Brugger et al., 2000; Ridsdale et al., 2006;Sun et al., 2005). Also, increased cholesterol content of ER membranes has been observed to inhibit SRP-dependent protein translocation into the ER due to a disruption of interactions between ribosomal-nascent chain complexes and the Sec61 PCC (Nilsson et al., 2001).
Consequently, the ER membrane is sensitive to cholesterol levels: Increased levels of the sterol can induce the unfolded protein response (UPR) stress response, and cause protein inactivation probably due to increased stiffening of the usually fl uid ER membranes (Davis and Poznansky, 1987; Li et al., 2004).
The lipid composition of membranes has also been implicated in regulating membrane insertion and translocation of proteins in prokaryotes. The emphasis in these studies has been on the roles of anionic lipids in bacterial inner
membranes. This membrane bilayer is rich in the non-bilayer lipids such as phosphatidylethanolamine (PE) that prefers to assemble into inverted non-bilayer structures and the anionic lipids phosphatidylglycerol (PG) and cardiolipin (Raetz, 1978). In several studies, both in biological and model membranes, anionic lipids have been shown to be important for signal peptide-lipid interactions (Batenburg et al., 1988; Kusters et al., 1991; Phoenix et al., 1993). Charged lipids have also been shown to enhance hydrophobic peptide binding and the formation of alpha-helical structures upon membrane association of the peptide (Liu and Deber, 1997;Ren et al., 1999). Similarly, the bacterial SecYEG complex is dependent on anionic and non-bilayer lipids for optimal translocation activity (van der Does et al., 2000).
Charged lipids have been implicated in targeting of presequences of polypeptides to the MOM in eukaryotes. Mitochondrial membranes are characterized by elevated levels of cardiolipin that is found mostly in the inner mitochondrial membrane (MIM), with low amounts detected in the outer membrane, concentrated at contact sites between the outer and inner membranes (Table 6) (Ardail et al., 1990; Hostetler and van den Bosch, 1972; Hovius et al., 1993).
In in vitro assays, some presequences show binding specifi city for cardiolipin (Ou et al., 1988; Rietveld et al., 1986), which induces the peptides to form α-helical structures (Goormaghtigh et al., 1989;
Leenhouts et al., 1994). Interestingly, the pro-apoptotic protein tBid, which does not have a presequence, has been observed to be specifi cally targeted to cardiolipin rich sites on the MOM (Lutter et al., 2000).
Direct membrane insertion of proteins has been widely studied in the outer envelope of chloroplasts in plants (Schleiff and Klosgen, 2001). Most proteins of
Discussion and conclusions
the outer envelope of chloroplasts are encoded by nuclear genes and are post-translationally translocated through the membrane. Most proteins of the outer envelope are translocated through the Toc and Tic PCCs (translocon in the outer/
inner envelope of chloroplasts) via transit-peptide mediated protein targeting that is analogous to SRP-dependent targeting at the ER membrane (Bruce, 2001; Jarvis and Soll, 2001). Membrane insertion of TA proteins into the outer chloroplast envelope does not require ATP, GTP, or protease sensitive factors, although these factors can increase the membrane insertion effi ciency (Schleiff and Klosgen, 2001). Interestingly, direct membrane insertion of several TA proteins has been shown to be temperature sensitive, and to depend on membrane lipid composition. The TA protein Toc34, for example, inserts preferentially into membranes containing non-bilayer lipids (Qbadou et al., 2003), while lowering the temperature reduces the membrane insertion eff iciency. The reduction in temperature induces a phase transition of the lipids from a gel to the more ordered liquid crystalline phase (Salomon et al., 1990; Schleiff and Klosgen, 2001). One can therefore conclude that the correct lipid environment is important for direct membrane insertion to occur: Membrane association and therefore targeting of hydrophobic peptides is enhanced by charged lipids at the MOM, while direct membrane insertion is possible in membranes of the appropriate fl uidity. A reduction in membrane fluidity induced either by cholesterol loading at the ER membrane or a decrease in temperature at the outer membrane of choroplasts, inhibits membrane insertion. Since, at least for b(5), no membrane proteins are required for membrane insertion, the
inhibition is likely to be due to changes in the nature of the TMD-lipid interactions.
It is also possible that a reduction in membrane fluidity increases the energy required to insert the TMD into the lipid bilayer and thus, the free energy liberated by membrane insertion under normal conditions is no longer suffi cient to drive membrane integration. Also, the results presented here and the work of others demonstrates that direct membrane insertion is a ubiquitous mechanism found in several distinct membranes:
The outer envelope of chloroplasts, the inner membrane of bacterial cells, and the mitochondria and ER membranes of eukaryotes. This mechanism requires, as a prerequisite, a sufficiently long TMD that can span the bilayer (Ren et al., 1999).
This domain plays an essential role in both membrane targeting and membrane insertion.
Discrepancies, however, are found in the requirements for energy and protease-sensitive protein activities.
Some TA proteins, such as b(5) and Bcl-2, do not require either factor for effi cient membrane insertion, while others such as synaptobrevin 2 do (I and II, (Abell et al., 2004;Kim et al., 1997). Thus, we studied further the effect of the TMD of b(5) on membrane insertion, and the energy and protein factors required (Brambillasca, et al, submitted). Indeed, direct insertion was observed to depend on the hydrophobcity of the TMD. When the hydrophobicity of the TMD of b(5)-Nglyc was increased, membrane insertion into liposomes was inhibited, and insertion into microsomes became ATP-dependent and trypsin-sensitive. This behaviour was also observed when the TMD of b(5) was replaced by that of synaptobrevin 2. Membrane insertion of synaptobrevin 2 has been shown to require
Discussion and conclusions
ATP and a protease sensitive factor.
These results are in keeping with data gathered from TA proteins of the outer envelope of chloroplasts where TMDs of low hydrophobicity can “spontaneously”
insert into membranes, whereas more hydrophobic domains show a dependence on energy and a protease sensitive factor for membrane insertion (Abell et al., 2004;Kutay et al., 1995b;Schleiff and Klosgen, 2001). A hypothesis arises in which the energy and protein factors are required to maintain the highly hydrophobic domains in a translocation-competent state.
Another question raised by our work on b(5)-Nglyc concerns the length of the amino-acids that could be spontaneously translocated across the membrane. The luminal fragment of b(5)-Nglyc is 28 amino acids long (I). As a criterion for TA proteins, the region downstream of the TMD has been proposed to have a maximum of 30 amino acids (Borgese et al., 2003b). Indeed, in plants, direct membrane insertion can translocate hydrophilic sequences of up to 30 amino acids across the bilayer (Schleiff and Klosgen, 2001). We have shown that protein sequences of up to 85 amino acids appended to the C-terminus b(5)-Nglyc could be translocated across microsomal membranes and protein-free lipsomes in vitro, in the absence of energy. The fragments studied differed in their net charge and protein fold. Even more surprisingly, in vivo, in yeast, fragments as long as 128 amino acids were effi ciently translocated independently of the SRP- and Sec61 PCC translocation pathway (Brambillasca et al, submitted). Once again, the importance of the TMD was demonstrated, as the more hydrophobic TMD of synaptobrevin 2 was unable to
support translocation of such long peptides across the protein free bilayers, in vitro.
Taken together, it appears that direct and “spontaneous” membrane insertion of TA proteins is a ubiquitous mechanism by which short TMDs can rapidly integrate into membrane bilayers. Target membranes are apparently recognized by their characteristic fl uidity. A decrease in membrane fl uidity, induced by decreasing temperature or increasing cholesterol content, inhibits the membrane insertion of TA proteins.
7.2 Targeting of C-tail anchored proteins to mitochondrial membranes Based on the criteria for MOM and ER targeting gathered from several studies (Figure 6), N-Bak with its moderately hydrophobic TMD flanked by positive charges would be predicted to be targeted to the MOM. As the TMD of N-Bak is moderately hydrophobic it should also be predicted to insert into the membrane (On the GES hydrophobicity scale, the TMDs of N-Bak, cytochrome b(5), synaptobrevin 2 and Bcl-2 are -1.7, -1.5, -2.6 and -1.7, respectively). Indeed, when expressed in yeast, N-Bak did associate with the MOM.
Surprisingly, however, N-Bak associated with membranes as a peripheral membrane protein. Despite this, the active protein induced statically signifi cant swelling of mitochondrial. In contrast, in sympathetic neurons extensive ER proliferation was observed, whereas mitochondria exhibited signs of MOM degradation.
One could infer from these results that for N-Bak to insert into mitochondrial membranes interactions with proteins of the Bcl-2 family may be required. It is probable that such interactions would be mediated by the BH3 domain of N-Bak.
As such interactions are not possible in
Discussion and conclusions
yeast, the TMD of N-Bak could remain buried within the tertiary structure and thus is unable to insert into the membrane.
As BH3-only proteins are believed to act as sentinels of specifi c cell damage, it is also possible that the inability of N-Bak to insert into yeast MOM refl ects the lack of a regulatory factor: in the absence of the proper “trigger” the TMD is not able to insert into the membrane. The possibility that N-Bak is unable to translocate across the membrane due to def iciencies in the protein import machinery of yeast mitochondria is highly unlikely as the mitochondrial import machinery is conserved and, in addition, Bcl-2 family proteins have not been shown to require these import machineries.
In sympathetic neurons, the observed ER proliferation could arise from the targeting of N-Bak to the ER and/or MOM membranes. As it is highly probable that in this cellular context the BH3 domain of
N-Bak is recognized by a Bcl-2 like protein, the TMD of N-Bak would be exposed and able to insert into a membrane. It would be of great interest to elucidate what membrane N-Bak is associated with in sympathetic neurons. That N-Bak containing an active BH3 domain was observed to induce mitochondrial swelling in yeast i.e. in a cellular context devoid of other Bcl-2 proteins is also interesting.
Could this be due to a mechanism of action of the BH3 domain not restricted to interactions with other Bcl-2 proteins?
More probably, the structure of the active protein is such that when associated with the membrane surface it induces a similar membrane proliferation, although to a much lesser degree, induced by bona-fi de TA proteins at the ER membrane (I, (Vergeres et al., 1993)). The mechanisms by which overexpression of TA proteins induces membrane proliferation are not understood.
Discussion and conclusions